One way that an
individual can protect their personal data is simply to not participate in a study. This is known as
the opt-out option. However, there are a
few considerations for this as a solution. Even if you decide to
opt-out of a study, it may still produce
results that affect you. For example, you may choose to opt-out of a study that
compares the heart disease diagnoses across a couple of people on the basis
that doing so may reveal a heart disease
diagnosis that causes your health
insurance premiums to rise. The study finds a correlation
between people who drink coffee and a higher
risk of heart disease, and your insurance company knows that you're
a coffee drinker, your rate may rise
even though you didn't personally
participate in the study. The benefits of
participation in the study may outweigh any
negative impact. For example, if you're paid a $100 to participate
in a study that results in your
health insurance rate rising by $10 per year, it will be more than 10 years
before you make a net loss. This may be a worthwhile
trade-off to you, particularly if you're rate may rise as a result of the study, even if you don't participate. The only way for the
opt-out option to work for every individual is for every individual
not to take part, which makes the whole
study pointless. The amount of variation
calls by adding noise is configurable through a
parameter called Epsilon. This value governs the amount of additional risks that
your personal data can be identified through rejecting
the opt-out option and participating in a study. The key thing is that it applies this privacy principle for everyone participating
in the study. A low Epsilon value provides
the most privacy at the expense of less accuracy
when aggregating the data. A higher Epsilon value
results in aggregations that are more true to the
actual data distribution, but in which the
individual contribution of a single individual to the aggregated value is
less obscured by noise.